python - Elegant and Efficient way retain date values as is without OOB error

Question

I have a dataframe which is like as shown below

df1_new = pd.DataFrame({'person_id': [1, 1, 3, 3, 5, 5],'obs_date': ['7/23/2377  12:00:00 AM', 'NA-NA-NA NA:NA:NA', 'NA-NA-NA NA:NA:NA', '7/27/2277  12:00:00 AM', '7/13/2077  12:00:00 AM', 'NA-NA-NA NA:NA:NA']})

As you can see few of my date values are out of bound values. However, I would still like to retain them as it is. Unfortunately, I couldn't due to OOB issue

I tried below

pd.to_datetime(df1_new['obs_date'], format='%m/%d/%Y %I:%M:%S %p', errors='coerce')

Is there any other efficient way to retain the date value as is but by changing the format alone? I am fine if it can be string column/datatype

I expect my output to be like as shown below.

updated try/except screenshot

Answer 1

You can convert values to datetimes and then to day Period for only possible format in pandas for represent out of bound values.

If omit it, then working with python datetimes objects, not with pandas datetimes (timestamps).

from datetime import datetime
def str2time(x):
    try:
        return pd.Period(datetime.strptime(x, '%m/%d/%Y %I:%M:%S %p'), 'D')
    except:
        return np.nan

df1_new['obs_date'] = df1_new['obs_date'].apply(str2time)
print(df1_new)
   person_id    obs_date
0          1  2377-07-23
1          1         NaT
2          3         NaT
3          3  2277-07-27
4          5  2077-07-13
5          5         NaT

print(df1_new['obs_date'].dtype)
period[D]

If possible multiple formats:

def str2time(x):
    try:
        #MM/DD/YYYY II:MM:SS pp like 7/23/2377  12:00:00 AM
        return pd.Period(datetime.strptime(x, '%m/%d/%Y %I:%M:%S %p'), 'D')
    except:
        try:
            #YYYY-MM-DD HH:MM:SS like 2377-07-23 00:00:00
            return pd.Period(datetime.strptime(x, '%Y-%m-%d %H:%M:%S'), 'D')
        except:
            return np.nan

df1_new['obs_date'] = df1_new['obs_date'].apply(str2time)